Method and system for providing content feed using answers from large language model

Abstract

The present invention relates to a method and system for providing a content feed using answers from a large language model. More specifically, the method for providing a content feed using answers from a large language model according to the present invention may comprise the steps of: receiving, from an answer generation system for generating the answers by using the large language model, a user query-response data set composed of a plurality of user queries and a plurality of answers acquired from the large language model for each of the plurality of user queries; generating a plurality of posts by using the user query-response data set; providing a content feed including the plurality of posts to a user terminal; and providing, to the user terminal, different pieces of information related to the user query-response data set on the basis of different types of user inputs to the user terminal.

Description

Method and system for providing content feeds using responses from large-scale language models

[0001] The present invention relates to a method and system for providing a content feed using answers from a large language model.

[0002] Recently, with the advancement of artificial intelligence technology, large language models (LLMs, e.g. Chat GPT) capable of natural conversations with people have emerged.

[0003] Unlike chatbots, which only provide limited, manually constructed answers, these large-scale language models are revolutionizing the artificial intelligence market by communicating naturally and human-like, and providing a variety of information, including information retrieval, question-answering, text summarization, and language translation, quickly and accurately.

[0004] As the utilization of large-scale language models increases, various services utilizing these models are being offered. For example, Korean Patent No. 10-2674954 provides a search service utilizing a large-scale language model. It determines whether search items are duplicated, preventing duplicate information from appearing in search results.

[0005] Furthermore, there is a need for methods to improve user convenience for large language models and provide an environment for exploring diverse information.

[0006] The present invention relates to a method and system for conveniently providing a user with an answer from a large language model.

[0007] More specifically, the present invention relates to a method and system for providing answers from a large language model via a content feed.

[0008] Furthermore, the present invention relates to a method and system for providing an environment in which a user can share answers obtained from a large language model with other users.

[0009] In order to solve the problem discussed above, a method for providing a content feed using answers from a large language model according to the present invention may include the steps of: receiving, from an answer generation system that generates answers using the large language model, a user question-response data set composed of a plurality of user queries and a plurality of answers obtained from the large language model for each of the plurality of user queries; generating a plurality of posts using the user question-response data set; providing a content feed including the plurality of posts to a user terminal; and providing different information related to the user question-response data set to the user terminal based on different types of user input to the user terminal.

[0010] Furthermore, in the step of generating the plurality of posts, the plurality of posts may be generated such that each of the plurality of posts includes one user query and one answer to one of the user queries among the plurality of user queries and the plurality of answers included in the user query-response data set.

[0011] Furthermore, in the step of generating the plurality of posts, a first post may be generated using a first user query and a first answer that is an answer to the first user query among the plurality of user queries and the plurality of answers included in the user question-response data set, and a second post may be generated using a second user query that is different from the first user query and a second answer that is an answer to the second user query.

[0012] Furthermore, in the step of providing the content feed to the user terminal, a plurality of graphic objects corresponding to each of the plurality of posts may be included, and the plurality of graphic objects may be provided to the user terminal in a manner of being arranged in a preset size, preset shape, and preset arrangement.

[0013] Furthermore, so that a user of the user terminal can view the plurality of user queries through the content feed, each of the plurality of graphic objects may include a specific user query included in a specific post corresponding to each of the plurality of graphic objects.

[0014] Furthermore, the user input of the different types includes a first type of user input and a second user input different from the first type, and in the step of providing the different information, when the user input of the first type is applied to the user terminal, the user input of the first type is processed as a scroll command for scrolling the content feed, and at least a portion of the content feed is updated based on the scroll command, and when the user input of the second type is applied to the user terminal, the user input of the second type is processed as a selection command for selecting a specific post from the content feed, and based on the selection command, an answer corresponding to the specific post is provided to the user terminal.

[0015] Furthermore, in the step of providing the different information, the content feed may be moved in one direction so that at least one different post, different from the post that was being output before the first type of user input was applied to the user terminal, is provided to the user terminal based on the scroll command.

[0016] Furthermore, based on the selection command, the content feed page including the content feed may disappear from the user terminal, and an answer page including an answer corresponding to the specific post may be provided to the user terminal.

[0017] Furthermore, on the above response page, based on user input applied to the user terminal, a response corresponding to the specific post and a response corresponding to a different post from the specific post can be continuously provided.

[0018] Furthermore, at least some of the plurality of posts provided through the content feed may be generated based on user queries made by users of user accounts logged into the user terminal and other users through the answer generation system.

[0019] Furthermore, a specific user query corresponding to the plurality of posts constituting the content feed and an answer corresponding to the specific user query are selected from the user query-response data set based on user information of a user account logged into the user terminal, and the user information may include information related to at least one of the user's age, residence, occupation, interests, and hobbies.

[0020] Meanwhile, a method and system for providing a content feed using answers from a large language model according to the present invention receive a user question-response data set composed of a plurality of user queries and a plurality of answers obtained from the large language model for each of the plurality of user queries from an answer generation system that generates answers using the large language model, generates a plurality of posts using the user question-response data set, provides a content feed including the plurality of posts to a user terminal, and provides different information related to the user question-response data set to the user terminal based on different types of user input to the user terminal.

[0021] Meanwhile, a program according to the present invention is a program that is executed by one or more processes in an electronic device and can be stored in a computer-readable recording medium, wherein the program may include instructions for performing the steps of: receiving, from an answer generation system that generates answers using a large language model, a user question-response data set comprising a plurality of user queries and a plurality of answers obtained from the large language model for each of the plurality of user queries; generating a plurality of posts using the user question-response data set; providing a content feed including the plurality of posts to a user terminal; and providing different information related to the user question-response data set to the user terminal based on different types of user input to the user terminal.

[0022] Furthermore, a method for providing a content feed using an answer from a large language model according to the present invention may include a step of receiving a user query, a step of inputting a prompt including the user query into the large language model to obtain an answer corresponding to the user query from the large language model, a step of providing the answer to the user terminal, a step of receiving a sharing request for the user query and the answer from the user terminal, and a step of providing a user query-response data set including the user query and the answer to a content feed providing system in response to the sharing request.

[0023] Furthermore, the system may further include a step of providing a content feed consisting of posts including the user query and the answer to users connected to the content feed provision system.

[0024] As described above, the method and system for providing a content feed using a response from a large language model according to the present invention can build a platform for sharing information obtained from a large language model by providing a content feed including a plurality of posts generated using a user query and a response from a large language model to the user query to a user terminal.

[0025] Furthermore, the method and system for providing a content feed using answers from a large language model according to the present invention can provide a user-friendly information search service by sharing information obtained from a large language model with multiple users, thereby enabling users who are not familiar with using large language models to obtain information that meets their needs.

[0026] In addition, the method and system for providing a content feed using answers from a large language model according to the present invention can improve user experience by increasing familiarity with the large language model and quickly providing customized information tailored to various interests by providing a content feed that includes the characteristics of a social network service (SNS) familiar to users.

[0027] FIG. 1 is a block diagram illustrating a content feed provision system and an answer generation system using answers from a large language model according to the present invention.

[0028] Figures 2a and 2b are conceptual diagrams for explaining a content feed provision process using answers from a large language model according to the present invention.

[0029] FIG. 3 is a flowchart illustrating a method for providing a content feed using a response from a large language model according to the present invention.

[0030] FIG. 4 is a conceptual diagram illustrating a process of generating a question-response set using answers obtained from a user question and answer generation system according to the present invention.

[0031] FIG. 5 is a conceptual diagram illustrating a process of generating multiple posts using multiple question-response sets according to the present invention and providing a content feed including the multiple posts to a user terminal.

[0032] FIGS. 6A to 6D are conceptual diagrams illustrating an embodiment of providing different information to a user terminal according to a user input according to the present invention.

[0033] Figure 7 is a conceptual diagram illustrating an embodiment of sharing and storing a specific post according to the present invention.

[0034] FIG. 8 is a conceptual diagram illustrating an embodiment of providing a content feed according to the present invention in a customized manner based on user information.

[0035] FIGS. 9A to 9C are conceptual diagrams illustrating an embodiment in which answers corresponding to a specific post are continuously provided based on user input according to the present invention.

[0036] Hereinafter, embodiments disclosed in this specification will be described in detail with reference to the attached drawings. Regardless of the drawing numbers, identical or similar components will be given the same reference numbers, and redundant descriptions thereof will be omitted. The suffixes "module" and "part" used for components in the following description are assigned or used interchangeably only for the convenience of writing the specification, and do not in themselves have distinct meanings or roles. In addition, when describing the embodiments disclosed in this specification, if it is determined that a specific description of a related known technology may obscure the gist of the embodiments disclosed in this specification, a detailed description thereof will be omitted. In addition, the attached drawings are only intended to facilitate easy understanding of the embodiments disclosed in this specification, and the technical ideas disclosed in this specification are not limited by the attached drawings, and should be understood to include all modifications, equivalents, and substitutes included in the spirit and technical scope of the present invention.

[0037] Terms that include ordinal numbers, such as first, second, etc., may be used to describe various components, but the components are not limited by these terms. These terms are used solely to distinguish one component from another.

[0038] When a component is referred to as being "connected" or "connected" to another component, it should be understood that it may be directly connected or connected to that other component, but that there may be other components intervening. Conversely, when a component is referred to as being "directly connected" or "connected" to another component, it should be understood that there are no other components intervening.

[0039] Singular expressions include plural expressions unless the context clearly indicates otherwise.

[0040] In this application, terms such as “include” or “have” are intended to specify the presence of a feature, number, step, operation, component, part or combination thereof described in the specification, but should be understood not to exclude in advance the possibility of the presence or addition of one or more other features, numbers, steps, operations, components, parts or combinations thereof.

[0041] The present invention relates to a method and system for providing a content feed using answers from a large language model.

[0042] Here, the “Large Language Model” refers to an artificial intelligence model that can learn a large amount of data, understand natural language, and generate it. The type of large language model used in the answer generation system according to the present invention is not limited, and various large language models can be used depending on the purpose or utilization. For example, the large language model of the present invention may include at least one of GPT (Generative Pre-trained Transformer), BERT (Bidirectional Encoder Representations from Transformers), LaMDA (Language Model for Dialogue Applications), PaLM (Pathways Language Model), Gemini, HyperCLOVA, and ExaONE.

[0043] Furthermore, a "content feed" may refer to a scrollable stream of content. More specifically, a content feed may refer to a continuous list of content, allowing users to easily consume the content. For example, a content feed may include a news feed, which corresponds to a stream of news articles, or a social media feed, which continuously lists people's photos, videos, and stories. In the present invention, answers from a large language model are provided to users through a content feed, and users can receive various answers obtained from the large language model without having to directly input queries into the large language model.

[0044] Hereinafter, a method and system for providing a content feed using a large language model will be described in more detail with reference to the attached drawings. Fig. 1 is a block diagram illustrating a content feed providing system and an answer generating system using answers from a large language model according to the present invention, and Figs. 2a and 2b are flowcharts for explaining a content feed providing method using answers from a large language model according to the present invention. Fig. 3 is a flowchart for explaining a content feed providing method using answers from a large language model according to the present invention, and Fig. 4 is a conceptual diagram for explaining a process of generating a question-answer set using answers obtained from a user question and answer generating system according to the present invention. Fig. 5 is a conceptual diagram for explaining a process of generating a plurality of posts using a plurality of question-answer sets according to the present invention and providing a content feed including the plurality of posts to a user terminal, and Figs. 6a to 6d are conceptual diagrams for explaining an embodiment of providing different information to a user terminal according to a user input according to the present invention. FIG. 7 is a conceptual diagram for explaining an embodiment of sharing and storing a specific post according to the present invention, FIG. 8 is a conceptual diagram for explaining an embodiment of providing a content feed according to the present invention in a customized manner according to user interests, and FIGS. 9a to 9c are conceptual diagrams for explaining an embodiment of providing answers corresponding to a specific post continuously based on user input according to the present invention.

[0045] As illustrated in FIG. 1, a content feed providing system using a large language model according to the present invention (hereinafter, referred to as “content feed providing system” and described as 100) may include at least one of an input unit (110), an output unit (120), a communication unit (130), a storage unit (140), and a control unit (150).

[0046] At this time, the content feed provision system (100) according to the present invention is not limited to the above-described components, and may further include components that perform the same or similar roles as the functions described in this specification.

[0047] The input unit (110) serves as a means for data input and may be configured in various ways. For example, the input unit (110) may be configured to receive user input. The input unit (110) may be configured to receive user input from a user terminal (10). Here, “receiving input” may mean receiving an input signal (or selection signal) corresponding to the user’s input based on the input being made by the user through the input unit configuration provided in the user terminal (10).

[0048] The input unit (110) may also be referred to as a user interface module. The input unit (110) may include a touch screen, a computer mouse, a keyboard, a keypad, a touch pad, a trackball, a joystick, a voice recognition module, or other similar devices. However, the present invention does not place any limitations on the type of input unit (110). Furthermore, the input unit (110) in the present invention does not necessarily refer to a hardware means, but may be understood as a channel for receiving input from a user. Here, the user input may include at least one of a document, text, an image (or video), and voice. For example, the input unit (110) may receive a user query from a specific user and may receive user input regarding a plurality of posts constituting a content feed. The input unit (110) may refer to an input means provided in the user terminal (10).

[0049] Next, the output unit (120) can be configured to output various information provided in the present invention. In the present invention, the output unit (120) can be understood to mean an output means (e.g., a display unit, a touch screen, a speaker, etc.) provided in the user terminal (120). In the present invention, outputting information (e.g., content feed) to the output unit (120) can be understood to mean outputting information to the display of the user terminal (10).

[0050] The communication unit (130) may be connected to at least one of the answer generation system (200), the user terminal (10), and the server (e.g., a central server, an external server, etc.) via a wireless or wired network, and may be configured to receive or transmit information necessary for providing a content feed.

[0051] The communication unit (130) can receive a user question-response data set from the answer generation system (200) to generate a content feed.

[0052] In the present invention, the answer generation system (200) may refer to a system that generates an answer using a large language model. In the database (DB) of such an answer generation system (200), a user query and a response from a large language model to the user query may be matched and stored as a data set. For example, assuming that a first answer (ex: “Tell me a Busan restaurant tour route”) is obtained from a large language model for a first user query (ex: “This is a Busan restaurant tour route. Songdo Sacheon Haemultang is a traditional seafood restaurant located near Songdo Beach”), the first answer and the first user query may be matched and stored as a user question-response data set in the answer generation system (200).

[0053] A user question-response data set may include a plurality of user queries and a data set in which answers from a large language model are matched for each of the plurality of user queries (e.g., a data set in which a first user query matches a first answer, a data set in which a second user query matches a second answer).

[0054] Meanwhile, the communication unit (130) may provide a content feed to the user terminal (10). This content feed may be generated based on a user question-response data set received from the answer generation system (200). Here, the user terminal (10) may include at least one of a mobile phone, a smart phone, a notebook computer, a laptop computer, a slate PC, a tablet PC, an ultrabook, a desktop computer, a digital broadcasting terminal, a personal digital assistant (PDA), a portable multimedia player (PMP), a navigation device, and a wearable device (e.g., a smartwatch, a smart glass, a head mounted display (HMD)).

[0055] Furthermore, the communication unit (130) can support various communication methods according to the communication standards of the communicating device. For example, the communication unit (130) may be configured to communicate with a communication target using at least one of WLAN (Wireless LAN), Wi-Fi (Wireless-Fidelity), Wi-Fi (Wireless Fidelity) Direct, DLNA (Digital Living Network Alliance), WiBro (Wireless Broadband), WiMAX (World Interoperability for Microwave Access), HSDPA (High Speed ​​Downlink Packet Access), HSUPA (High Speed ​​Uplink Packet Access), LTE (Long Term Evolution), LTE-A (Long Term Evolution-Advanced), 5G (5th Generation Mobile Telecommunication), Bluetooth™, RFID (Radio Frequency Identification), Infrared Data Association (IrDA), UWB (Ultra-Wideband), ZigBee, NFC (Near Field Communication), Wi-Fi Direct, and Wireless USB (Wireless Universal Serial Bus) technologies.

[0056] Next, the storage unit (140) may be configured to store various information related to the present invention. In the present invention, the storage unit (140) may be provided within the content feed provision system (100) itself, or at least a portion of the storage unit may be provided in an external cloud server (or external cloud storage). That is, in the present invention, the storage unit (140) may be sufficient as long as it stores information necessary for the content feed provision system (100), and it can be understood that there are no restrictions on physical space. The storage unit (140) may store and exist various information necessary for providing content feeds using responses from a large language model. More specifically, the storage unit (140) may store i) a user question-response data set received from the answer generation system (200), ii) a plurality of posts generated using the user question-response data set, iii) information related to the plurality of posts (e.g., at least one of text, image, number of views, number of likes, number of saves, and comments), and iv) user information of a user account (e.g., information related to at least one of the user's age, place of residence, occupation, interests, and hobbies).

[0057] Here, the user information may include at least one of the user's activity history information (e.g., user's interest content (e.g., user's like information, user's comment information, user's selected (or used) content information, user's search history information, etc.), user's preference information (e.g., may be obtained through analysis of activity data), recommended content for the user, and content exposure history (e.g., time-based record data on content placement).

[0058] Furthermore, the storage unit (140) may store content metadata for each of a plurality of posts. Here, "metadata" may refer to data describing a specific post. More specifically, the metadata may include at least one of a user question-and-answer data set related to a specific post, the upload time of the specific post, location information, hashtag and post topic information, and user information regarding the user who received the specific post.

[0059] The storage unit (140) may store user terminal information including at least one of the device information of the user terminal, the operating system of the user terminal, the browser type, and the app usage history. Furthermore, the storage unit (140) may store user session data including at least one of the user login time, usage time, and time spent on the feed. Furthermore, the storage unit (140) may store information regarding advertisements displayed in a content feed provided to the user terminal. Specifically, the storage unit (140) may store at least one of the number of user clicks, likes, and shares for the advertisement, and a purchase history for a product corresponding to the advertisement displayed in the feed.

[0060] Furthermore, the storage unit (140) may store specific posts selected by specific user input, group by group. Specifically, the storage unit (140) may store a specific post group related to a specific post based on whether a specific user has stored the specific post, and the specific post may be stored under the specific group.

[0061] Next, the control unit (150) can perform overall control related to the provision of content feeds. The control unit (150) can process signals, data, information, etc. input or output through the components discussed above, or provide or process appropriate information or functions to the user.

[0062] As illustrated in FIG. 2a, the control unit (150) can provide a content feed service based on linkage with the database (LLM answer database, 210) of the answer generation system (200).

[0063] As described above, a user question-response data set in which a user query and an answer obtained from a large language model are matched may be stored and exist in the database (210) of the answer generation system (200). As illustrated in (a) of Fig. 2b, a user question-response data set (210a) may be stored and exist in the database (210) of the answer generation system (200). The user question-response data set (210a) may include at least a portion of user information corresponding to a query (e.g., information on a user who input a user query, 211), a user query (212), and an answer (213) of a large language model to the user query.

[0064] The control unit (150) can provide a content feed service that allows users to view answers from large-scale language models and create, share, and store posts constituting the content feed, through a content feed using a user question-response data set stored in the answer generation system (200). For example, as illustrated in (b) of FIG. 2B, the control unit (150) can match user information (141) of a specific user account, content feed information (142) provided to a specific user, content post information (143) generated based on an answer to a specific user's question, and recommended post information (e.g., like information, 144) with each other and manage them in a storage unit (or database, 140).

[0065] Meanwhile, the content feed provision system (100) may include one or more processors, which may include one or more general-purpose processors and / or one or more special-purpose processors (e.g., digital signal processors, tensor processing units (TPUs), graphics processing units (GPUs), neural network processing units (NPUs), application-specific integrated circuits, application-specific integrated circuits (ASICs), etc.). The one or more processors may be configured to execute instructions stored (or included) in the storage unit (140), computer-readable instructions, and / or other instructions described herein. The content feed provision system may enable the memory and at least one processor to cooperate to perform data processing as described below. The processor may perform a series of operations and data processing using data and information stored in the memory. Here, the “memory” may be a component of the storage unit (140), and the “processor” may be used interchangeably with the control unit (150).

[0066] Below, we describe in more detail the content feed service provided using responses from large language models.

[0067] In the present invention, a process of receiving a user question-response data set composed of a plurality of user queries and a plurality of answers obtained from a large language model for each of the plurality of user queries may be performed from an answer generation system (200) that generates answers using a large language model (S310, see FIG. 3).

[0068] The control unit (150) may receive data including answers obtained from the large language model from the answer generation system (200) to provide a content feed using the answers from the large language model. In this case, the control unit (150) may receive both the answers from the large language model and a user query that matches the user query corresponding to the answers from the answer generation system.

[0069] As illustrated in FIG. 4, the answer generation system (200) may store and exist multiple data sets (441, 442, 443) in which multiple user queries (411, 412, 413) and answers (431, 432, 433) of a large language model for each of the multiple user queries are matched with each other. For example, let's assume that for a first user query (ex: "Tell me a Busan restaurant tour route"), a first answer (ex: "This is a Busan restaurant tour route. Songdo Sacheon Haemultang is a traditional seafood restaurant located near Songdo Beach") is obtained from a large language model, and for a second user query (ex: "What is the best match for ENFPs?"), a second answer (ex: "The best match for ENFPs is ENFJ") is obtained from a large language model. The answer generation system (200) may store and exist a first question-response set (441) in which the first user query (411) and the first answer (431) are matched, and a second question-response set (442) in which the second user query (412) and the second answer (432) are matched.

[0070] The control unit (150) can receive a user question-response data set (440) including a first question-response set (441) and a second question-response set (442) from the answer generation system (200).

[0071] Meanwhile, in the present invention, a large-scale language model of the answer generation system (200) can be used to obtain an answer to a user query received from a user terminal (10). More specifically, the control unit (150) can obtain an answer by using the user query (e.g., “Recommend a date course for a couple who have been dating for a week”) as input to the large-scale language model of the answer generation system (200) based on receiving the user query from the user terminal (10). The control unit (150) can generate a prompt (420) corresponding to the user query and process the generated prompt as input to the large-scale language model. In this case, the user query-response data set (440) may further include the prompt (420). More specifically, the control unit (150) can generate a plurality of prompts (420) corresponding to a plurality of user queries (410), respectively. For example, the control unit (150) can analyze a user query input into the user terminal (10) and extract key elements (e.g., keywords) of the user query. Based on the analysis result of the user query, the control unit (150) can generate a prompt corresponding to the user query and process the prompt as input to a large language model of the answer generation system (200) to obtain an answer. Here, the answer generation system (200) includes at least one large language model and can generate an answer to the input prompt using the large language model.

[0072] Meanwhile, in the present invention, a process of generating multiple posts can be performed using a user question-response data set (S320, see FIG. 3).

[0073] The control unit (150) can generate multiple posts for providing a content feed by using multiple user queries included in a user question-response data set and answers from a large language model corresponding to each of the multiple user queries.

[0074] The control unit (150) may generate the plurality of posts so that each post includes one user query and one answer to one of the plurality of user queries and answers included in the question-response data set. That is, the control unit (150) may generate at least one content post using a single data set comprising a user query and an answer to the user query.

[0075] As illustrated in FIG. 5, the control unit (150) can generate a first post (510) using a first user question-response set (441) composed of a first user query (411) and a first answer (431) to the first user query. In addition, the control unit (150) can generate a second post (520) using a second user question-response set (442) composed of a second user query (412) that is different from the first user query (411) and a second answer (432) that is an answer to the second user query.

[0076] Furthermore, the present invention may provide a content feed containing multiple posts to a user terminal (S330, see FIG. 3). The control unit (150) may provide a content feed containing posts generated using a user query and a large language model's response to the user query to the user terminal (10).

[0077] Here, "content feed" can refer to a scrollable stream of content. More specifically, a content feed can refer to a continuous display of content, allowing users to easily consume it. For example, a content feed can include a news feed, which represents a stream of news articles, or a social media feed, which presents a continuous display of people's photos, videos, and stories.

[0078] The control unit (150) can generate a plurality of graphic objects (510a, 520a) corresponding to each of a plurality of posts (510, 520). In addition, the control unit (150) can provide a content feed (500) including a plurality of graphic objects (510a, 520a) to the user terminal (10).

[0079] The control unit (150) can control a plurality of graphic objects (510a, 520a) to be displayed in a list on the display of the user terminal (10). For example, as illustrated in FIG. 5, the control unit (150) can generate each of the plurality of graphic objects (510a, 520a) in a preset size and a preset shape. In addition, the control unit (150) can arrange the plurality of graphic objects (510a, 520a) in a preset arrangement and output them on the display of the user terminal (10).

[0080] At this time, each of the plurality of graphic objects (510a, 520a) may include at least one of text and images related to a specific user query included in a specific post corresponding to each of the plurality of graphic objects.

[0081] The control unit (150) can include user queries included in multiple posts in each of multiple graphic objects so that a user of the user terminal (10) can view multiple user queries through the content feed (500).

[0082] For example, the control unit (150) may create a graphic object (510a) corresponding to the first post so that the first user query (ex: “Tell me a Busan restaurant travel route”, 411) is included, and a graphic object (520a) corresponding to the second post so that the second user query (ex: “What is the best compatibility for ENFPs?”, 412) is included.

[0083] Meanwhile, in the present invention, a process of providing different information related to a user question-response data set to a user terminal based on different types of user input to the user terminal may be performed (S340, see FIG. 3).

[0084] In the present invention, different types of user input may include a first type of user input and a second type of user input different from the first type. The control unit (150) may monitor user input on the user terminal (10) while the content feed (500) is provided to the user terminal (10). In addition, the control unit (150) may perform different data processing based on whether either the first type of user input or the second type of user input is applied to the user terminal (10).

[0085] As illustrated in (a) of FIG. 6A, the control unit (150) can process the first type of user input (611) into a scroll command for the content feed (500) based on the first type of user input applied to the user terminal (10). In response to the scroll command for the content feed (500), the control unit (150) can update at least a portion of the content feed (500) and provide the user terminal (10) with a content feed including a new plurality of posts (600), as illustrated in (b) of FIG. 6A.

[0086] The control unit (150) can update the content feed (500) by moving the content feed (500) in one direction (611) to include at least one post that is different from the plurality of posts that were being output before the first type of user input was applied to the user terminal (10), based on a scroll command for the content feed.

[0087] For example, let us assume that a first type of user input corresponding to a scroll command for the content feed (500) is applied while a content feed (500) including a graphic object (510a) corresponding to a first post (510) and a graphic object (520a) corresponding to a second post (520) is being output to a user terminal (10). Based on the scroll command for the content feed, the control unit (150) can control the user terminal (10) to cause the graphic objects (510a, 520a) corresponding to at least one of the first post (510) and the second post (520) to disappear and cause a graphic object (530a) corresponding to a new third post to be output.

[0088] The control unit (150) can continuously update the content feed to include new posts based on scroll commands applied to the content feed. Through infinite scrolling of the content feed, users can continuously receive a variety of posts generated based on responses from the large language model.

[0089] Meanwhile, as illustrated in FIG. 6b, the control unit (150) may process the second type of user input (620) different from the first type into a selection command for selecting a specific post from among a plurality of posts constituting the content feed, based on the fact that the second type of user input (620) is applied to the user terminal (10). In addition, the control unit (150) may provide the user terminal with a response page (630) including a specific user query (631) corresponding to the selected specific post and a response (632) to the specific user query, as illustrated in FIG. 6c, according to the selection command for selecting the specific post. At this time, the content feed page (1000) including the content feed may disappear from the user terminal (10), and the response page (630) may be output to the user terminal (10).

[0090] For example, let's assume that a graphic object (510) including a first user query (e.g., "Tell me about a Busan restaurant travel route") is selected according to a second type of user input. In response to the user input, the control unit (150) may provide a first post including the first user query and a response from a large language model to the first user query to the user terminal (10). In this case, the control unit (150) may control the content feeds previously output to the user terminal (10) to all disappear, and output a response page (630) including the first post.

[0091] Meanwhile, on the answer page (630), answers corresponding to a specific post and answers corresponding to a specific post and other posts can be provided in succession based on user input applied to the user terminal (10).

[0092] As illustrated in FIG. 6D, the control unit (150) can monitor a first type of user input for the user terminal (10) while an answer page (630) including a first post is output on the user terminal (10). The control unit (150) can process the first type of user input for the answer page (630) as a scroll command for the answer page (630). In response to the scroll command for the answer page (630), the control unit (150) can control a second post (640) different from the first post to be output on the answer page (630) in succession to the first post. For example, the control unit (150) can provide at least one of a second user query (or a graphic object corresponding to the second user query) and the second answer to the user terminal (10) in succession to the first user query and the first answer.

[0093] Meanwhile, the present invention can provide a service in which a user directly queries a large language model, generates a content feed using the obtained answers, and shares the generated content feed with other users.

[0094] Hereinafter, the content feed system can be described as providing a service that generates a content feed using responses from a large language model. The content provision system can also be described as providing a service that shares content feeds with other users. For convenience, the present invention distinguishes between the content feed generation system and the content feed provision system. However, the content generation feed system may be a component of the content provision system, and the content feed generation system and the content feed provision system may be used interchangeably.

[0095] As illustrated in FIG. 7, the content creation system can receive a user query (e.g., “Make a plan for living a new life”, 711) from a user terminal (10). Based on receiving the user query, the content creation system can generate a prompt including the user query and input it into a large language model. When the content creation system obtains an answer from the large language model, the content creation system can provide the answer (712) to the user terminal (10) so that the user can check the answer. When the answer (712) to the user query is provided to the user terminal (10), the content creation system can receive a sharing request for the user query (711) and answer (712) from the user terminal (10). For example, the control unit (150) can output a sharing icon linked to a user query and answer sharing function to the user terminal (10). The content creation system can process that a sharing request for the user query and answer has been received from the user terminal (10) based on a user input selecting a sharing icon.

[0096] The content creation system can, in response to a share request, provide a user question-response data set including the user question and the answer to the content feed providing system so that posts including the user question and the answer are provided to other users as a content feed.

[0097] The content feed provision system can generate posts using a user query-response set provided by the content creation system. Furthermore, the content feed provision system can provide a content feed containing shared posts to users accessing the content feed provision system.

[0098] Meanwhile, the present invention can provide a storage service that stores posts included in a content feed. Here, "post storage" can be understood as matching posts with user accounts and storing and managing them in a storage unit (or database). For example, a user can save useful information among multiple posts included in a content feed and later check the stored posts.

[0099] In this way, a user accessing the system according to the invention can share a user question-and-response set, consisting of a user query entered by the user and a corresponding answer to the user query, with other users via a content feed. Furthermore, users can store and manage at least one post included in the content feed. For convenience of explanation, a user who shares his or her user query and answer with other users will be referred to as a "sharing user," and a user who saves the post to his or her user account will be referred to as a "saving user."

[0100] As illustrated in FIG. 7, the control unit (150) can receive (or input) a user query (711) of a sharing user (710) from a user terminal (10). The control unit (150) can generate a prompt corresponding to the user query (711) and input the prompt into a large language model included in the answer generation system (200). Furthermore, the control unit (150) can obtain an answer (712) corresponding to the user query from the large language model and provide the answer (712) to the user terminal (10). The control unit (150) can receive a sharing request for the user query (711) and the answer (712) corresponding to the user query from the user terminal (10). In response to the sharing request, the control unit (150) can generate a sharing post including the user query (711) and the answer (712) corresponding to the user query. And the control unit (150) can provide a content feed including shared posts to another user's terminal.

[0101] The control unit (150) may receive a storage request for a shared post from a user terminal of a storage user (720) who has received a content feed including shared posts and wishes to store the shared posts. In response to the storage request, the control unit (150) may store the shared posts by matching them to the user account of the storage user (720). Different posts may be stored by matching them to each user account. In order to store and manage multiple shared posts by matching them to user accounts, the control unit (150) may create a post group (730) corresponding to the user account and store the shared posts under the post group (730).

[0102] Meanwhile, the present invention can provide a content feed tailored to the user. The control unit (150) can select at least some of the multiple posts generated based on a user question-and-response data set and provide them as a content feed to the user terminal (10). Various methods may be used to select the posts included in the content feed.

[0103] The control unit (150) can randomly select posts from among a plurality of posts to form a content feed to be provided to the user terminal (10). For example, if the user account on the user terminal (10) is not logged in, or if user information and user activity data of a logged-in user account do not exist, the control unit (150) can provide the user terminal (10) with a content feed (500) including a plurality of randomly selected posts.

[0104] Furthermore, the control unit (150) may include posts containing queries from users other than the user of the user account logged into the user terminal in the content feed.

[0105] As previously described, the user question-response dataset may further include user information corresponding to the user query. For example, if a first user query is received from a first user account, the user question-response dataset may contain matching user information for the first user account, the first user query, and the response from a large language model to the first user query.

[0106] The control unit (150) may provide a first post containing a first user query and answer matched to the first user account as a content feed to a terminal logged in with a second user account different from the first user account. Alternatively, the control unit (150) may restrict the first post from being provided to the first user account as a content feed to a terminal logged in with the first user account.

[0107] Furthermore, the control unit (150) can select posts to be included in the content feed from among multiple posts based on preset feed exposure criteria. In this case, the control unit (150) can specify posts to be included in the content feed based on user information of a user account logged into the user terminal. Here, the user information of the user account may include information regarding at least one of the user's age, residential area, occupation, interests, hobbies, user queries, posts of interest, and posts viewed.

[0108] The control unit (150) can identify the user's preferred questions and preferred responses from a user question-response data set based on user information. Furthermore, the control unit (150) can identify posts containing at least one of the user's preferred questions and preferred responses as posts to be included in a content feed.

[0109] For example, the control unit (150) may select a plurality of posts predicted to have high user interest based on user information. Here, "user interest" may refer to the degree of interest the user shows in a specific post. The control unit (150) may extract keywords included in user information and user activity data to provide a user-customized content feed to the user terminal. Furthermore, the control unit (150) may select recommended posts corresponding to the keywords from among the plurality of posts and provide a content feed including the recommended posts to the user terminal (10).

[0110] For another example, the control unit (150) may set a specific post as a recommended post if at least one of the popularity, number of saves, and number of views for the specific post included in the plurality of posts satisfies a preset criterion. Here, "popularity" is an indicator related to the number of likes received from multiple users for the specific post, and it can be understood that the more likes a post has, the higher the popularity. In this case, the preset criteria for popularity, number of saves, and number of views may vary depending on user settings.

[0111] As another example, the control unit (150) may select a plurality of posts including the recommended posts predicted to be of high user interest by using user activity data including at least one of the user's i) like information, ii) comment writing, iii) content selection information, and iv) search history information. Furthermore, the control unit (150) may provide the user-customized content feed to the user terminal (10).

[0112] Meanwhile, the control unit (150) can extract keywords of interest to the user based on user information. Furthermore, the control unit (150) can use the keywords of interest to specify posts to be included in the user's content feed.

[0113] As illustrated in FIG. 8, a database (140a) may store and exist a plurality of keyword groups (810, 820, 830) corresponding to each of a plurality of keywords. In the plurality of keyword groups (810, 820, 830), user question-response data sets or posts corresponding to each keyword may be matched and exist. For example, a first keyword group (810) matched with the keyword “age” may include posts related to age. Furthermore, a keyword group may include at least one keyword of a lower concept corresponding to a keyword of a higher concept, and posts corresponding to each keyword of the lower concept may be included. For example, the first keyword group (810) may include a post (813) matched with “youth (811)”, which is a first lower keyword of “age,” and a post (814) matched with “middle-aged (812)”, which is a second lower keyword of “age.”

[0114] The control unit (150) may provide a content feed (820) containing user-customized posts to a user terminal logged into a specific user account based on a keyword group. Specifically, the control unit (150) may extract keywords from user-related information, such as age, gender, and hobbies included in specific user information, to predict the user's level of interest, and specify a keyword group pre-stored in a database (140a) corresponding to the keywords.

[0115] For example, the plurality of keyword groups (800) previously stored in the database (140a) may include at least one of a first keyword group (810), a second keyword group (820), and a third keyword group (830), and each keyword group may store a plurality of keywords corresponding to various keywords such as age, gender, and hobbies. Specifically, the first keyword group stored in the database (140a) of the storage unit (140) may include a plurality of keywords according to various ages such as “youth” and “middle-aged,” and based on the matching of the plurality of keywords with keywords extracted from user information, it may be predicted that the user’s interest in recommended posts corresponding to the plurality of keywords will be high. Furthermore, the control unit (150) may select different recommended posts (813, 814) respectively corresponding to specific keywords, depending on the user’s interest. As an example, the control unit (150) may predict that the user has a high level of interest in the keyword “youth” based on the fact that the keyword included in specific user information is “youth”, and may provide a content feed (820) including recommended posts (813) corresponding to the keyword “youth.”

[0116] Meanwhile, the control unit (150) may select a plurality of posts including user-customized posts predicted to have high user interest by using user activity data including at least one of the user's i) like information, ii) comment writing information, iii) content selection information, and iv) search history information. Specifically, the control unit (150) may extract the topic and keywords of a specific post for which a "like" indicating a preference for a specific post has been entered, and classify the extracted topics and keywords according to a keyword group stored in the database (140a).

[0117] Furthermore, the control unit (150) can predict a high user interest in a specific topic and keyword based on the extracted like information, and provide a content feed including recommended posts corresponding to the keyword group to the user terminal (10) according to the user interest.

[0118] In addition, the control unit (150) can extract topics and keywords from user comment writing information entered for a specific post among the user activity data, and classify them according to keyword groups stored in the database (140a). For example, the control unit (150) can analyze the content of a user comment for a specific post entered into the user terminal (10), extract core topics and keywords, and classify them based on keyword groups stored in the database (140a).

[0119] In addition, the control unit (150) can extract keywords associated with at least one of specific post information and search history information selected according to user input from among multiple posts constituting the content feed in the above manner, and classify them based on a keyword group stored in the database (140a).

[0120] Furthermore, the control unit (150) can provide a content feed including posts associated with each classified topic and keyword to the user terminal (10) based on the keyword group.

[0121] Meanwhile, the control unit (150) can update at least some of the plurality of posts included in the content feed with other posts based on a scroll command being applied to the content feed while the content feed is being provided to the user terminal (10).

[0122] The control unit (150) can specify new posts to be included in the updated content feed. The control unit (150) can update the content feed by selecting at least some of the posts not included in the content feed among the multiple posts generated based on the user question-and-response data set.

[0123] In this case, the control unit (150) can select posts to be included in a content feed that is randomly updated among multiple posts, or select posts based on user information. That is, the control unit (150) can update the content feed based on the content feed exposure criteria described above.

[0124] The control unit (150) may change the user interest criteria for a plurality of posts included in the content feed based on the continuous processing of a scroll command for the content feed (500) by continuously inputting a first type of user input to the user terminal (10). Specifically, the control unit (150) may update the content feed to include posts with a higher user interest than before by using at least one of user information and user activity data based on the fact that only the first type of user input corresponding to the scroll command for the content feed is input. Here, “higher interest than before” may mean that there are a large number of specific keywords that match a plurality of keywords included in the post with keywords extracted from the user information.

[0125] Furthermore, the control unit (150) may adjust user interest in topics and keywords corresponding to the selected specific posts based on the user input (620) of the second method of the first method for selecting specific posts applied to the user terminal (10) provided with the content feed. Specifically, the control unit (150) may adjust user interest in keywords associated with the selected specific posts so that they are adjusted upward when a selection command for a specific post, rather than a scroll command, is performed. In addition, the control unit (150) may continuously provide posts corresponding to keywords for which user interest has been increased to the user terminal. In other words, the control unit (150) may provide the user terminal (10) with a user-customized content feed that includes posts similar to topics and keywords corresponding to the selected specific posts.

[0126] Meanwhile, when a response page (630) for a specific post selected by a second type of user input (620) is being output on the user terminal (10), if a first type of user input corresponding to a scroll command for the response page (630) is applied, the control unit (150) can provide a post different from the specific post to the user terminal (10) in succession to the specific post. In this case, the control unit (150) can provide another post whose keyword corresponds to the specific post. For example, the control unit (150) can control another post matching the keyword “Busan” to be output in succession to the specific post based on the selection of a specific post matching the keyword “Busan.” In addition, the control unit (150) may predict that the user interest in the selected specific post and the other post (640) is high based on the second type of user input (620) applied to the user terminal (10) for the specific post and the other post (640), and output at least one of a content feed page (1000) and a response page (631) including a plurality of posts related to the other post (640) to the user terminal (10).

[0127] Meanwhile, in response to a response from a large language model to a user query included in a plurality of posts constituting a content feed of a content feed providing system according to the present invention, an additional user query may be input to a user terminal (10).

[0128] As described above, when a specific post is selected from among a plurality of posts constituting a content feed provided to a user terminal (10) by a second type of user input, the control unit (150) can provide a response page including a user query corresponding to the specific post and a response to the user query to the user terminal (10).

[0129] Furthermore, the control unit (150) may receive an additional user query following the answer to the user query from the user terminal (10) in order to provide additional information related to the answer to the user query included in the answer page.

[0130] As illustrated in FIG. 9a, when a post including a user query such as “Tell me about a Busan restaurant travel route” and an answer to the user query is selected from among a plurality of posts constituting a content feed provided to a user terminal (10) according to a second type of user input, the control unit (150) can provide an answer page (900) corresponding to the post.

[0131] Furthermore, the control unit (150) may activate a function for inputting an additional user query so as to provide additional information related to an answer (910) to a user query such as “Tell me about a Busan restaurant travel route” included in the post. As an example, as illustrated in FIG. 9B, the control unit (150) may create an input space for inputting an additional user query (920) for the answer (910) based on the selection of the “Continue conversation with prompt” button (901) provided on the answer page (900) by user input, and may receive the additional user query (920) in the user terminal (10).

[0132] Furthermore, as illustrated in FIG. 9c, the control unit (150) can use the additional user query (920) to generate an additional prompt and input the additional prompt into the answer generation system (200). The control unit (150) can receive the additional user query-response set from the answer generation system (200) and output the answer (930) of the large language model for the additional user query (920) on the answer page (900) of the user terminal (10).

[0133] As described above, the method and system for providing a content feed using a response from a large language model according to the present invention can build a platform for sharing information obtained from a large language model by providing a content feed including a plurality of posts generated using a user query and a response from a large language model to the user query to a user terminal.

[0134] Furthermore, the method and system for providing a content feed using answers from a large language model according to the present invention can provide a user-friendly information search service by sharing information obtained from a large language model with multiple users, thereby enabling users who are not familiar with using large language models to obtain information that meets their needs.

[0135] In addition, the method and system for providing a content feed using answers from a large language model according to the present invention can improve user experience by increasing familiarity with the large language model and quickly providing customized information tailored to various interests by providing a content feed that includes the characteristics of a social network service (SNS) familiar to users.

[0136] Meanwhile, computer-readable media include all types of recording devices that store data that can be read by a computer system. Examples of computer-readable media include hard disk drives (HDDs), solid-state disk drives (SSDs), silicon disk drives (SDDs), ROMs, RAMs, CD-ROMs, magnetic tapes, floppy disks, and optical data storage devices.

[0137] Furthermore, the computer-readable medium may include a storage device and may be a server or cloud storage device accessible via communication. In this case, the computer may download the program according to the present invention from the server or cloud storage device via wired or wireless communication.

[0138] Furthermore, in the present invention, the computer described above is an electronic device equipped with a processor, i.e., a CPU (Central Processing Unit), and there is no particular limitation on its type.

[0139] Meanwhile, the above detailed description should not be construed as limiting in all respects and should be considered illustrative. The scope of the present invention should be determined by a reasonable interpretation of the appended claims, and all modifications within the equivalent scope of the present invention are intended to be included within the scope of the present invention.

Claims

1. A method for providing a content feed using answers from a large language model, A step of receiving a user question-response data set, comprising a plurality of user queries and a plurality of answers obtained from the large language model for each of the plurality of user queries, from an answer generation system that generates answers using the large language model; A step of generating multiple posts using the above user question-response data set; A step of providing a content feed including the plurality of posts to a user terminal; and A method for providing a content feed using answers from a large language model, characterized in that it includes a step of providing different information related to the user query response data set to the user terminal based on different types of user input to the user terminal.

2. In paragraph 1, In the step of creating the above multiple posts, For each of the above multiple posts, Among the multiple user queries and the multiple answers included in the above user query-response data set, A method for providing a content feed using answers from a large language model, characterized in that the plurality of posts are generated to include one user query and an answer to the one user query.

3. In paragraph 2, In the step of creating the above multiple posts, Among the multiple user queries and the multiple answers included in the above user query-response data set, Create a first post using a first user query and a first answer which is a response to the first user query, A method for providing a content feed using answers from a large language model, characterized in that a second post is created using a second user query different from the first user query and a second answer that is an answer to the second user query.

4. In paragraph 3, In the step of providing the above content feed to the user terminal, Contains multiple graphic objects corresponding to each of the above multiple posts, A method for providing a content feed using a response from a large language model, characterized in that the plurality of graphic objects are arranged in a preset size, preset shape, and preset arrangement and provided to the user terminal.

5. In paragraph 4, So that the user of the user terminal can view the plurality of user queries through the content feed, A method for providing a content feed using answers from a large language model, characterized in that each of the plurality of graphic objects includes a specific user query included in a specific post corresponding to each of the plurality of graphic objects.

6. In paragraph 1, The above different types of user input include a first type of user input and a second type of user input different from the first type, In the step of providing the above different information, When the user input of the first type is applied to the user terminal, the user input of the first type is processed as a scroll command for scrolling the content feed, Based on the above scroll command, update at least a portion of the above content feed, When the second type of user input is applied to the user terminal, the second type of user input is processed as a selection command for selecting a specific post from the content feed, A method for providing a content feed using an answer from a large language model, characterized in that an answer corresponding to the specific post is provided to the user terminal based on the above selection command.

7. In paragraph 6, In the step of providing the above different information, A method for providing a content feed using an answer from a large language model, characterized in that the content feed is moved in one direction so that at least one post different from the post being output before the first type of user input is applied to the user terminal is provided based on the scroll command.

8. In paragraph 6, Based on the above selection command, the content feed page including the content feed disappears from the user terminal, A method for providing a content feed using answers from a large language model, characterized in that an answer page including an answer corresponding to the specific post is provided to the user terminal.

9. In paragraph 8, A method for providing a content feed using answers from a large language model, characterized in that, on the above answer page, answers corresponding to the specific post and answers corresponding to posts other than the specific post are continuously provided based on user input applied to the user terminal.

10. In paragraph 1, At least some of the plurality of posts provided through the above content feed, A method for providing a content feed using answers from a large language model, characterized in that the answers are generated based on user queries posed by users of user accounts logged into the user terminal and other users through the answer generation system.

11. In paragraph 1, A method for providing a content feed using answers from a large language model, characterized in that a specific user query corresponding to the plurality of posts constituting the content feed and an answer corresponding to the specific user query are randomly selected from the user question-response data set.

12. In paragraph 1, A specific user query corresponding to the plurality of posts constituting the above content feed and an answer corresponding to the specific user query, Based on the user information of the user account logged in to the user terminal, selected from the user question-response data set, The above user information is, A method for providing a content feed using answers from a large language model, characterized in that the answers include information related to at least one of the user's age, place of residence, occupation, interests, and hobbies.

13. In a system that provides content feeds using responses from a large language model, The above system, Receive a user question-response data set, which comprises a plurality of user queries and a plurality of answers obtained from the large language model for each of the plurality of user queries, from an answer generation system that generates answers using the large language model; Using the above user question-response data set, multiple posts are generated, Providing a content feed including the above multiple posts to the user terminal, A content feed providing system using answers from a large language model, characterized in that it provides different information related to the user query response data set to the user terminal based on different types of user input to the user terminal.

14. A program that is executed by one or more processes in an electronic device and can be stored in a computer-readable recording medium, The above program is, A step of receiving a user question-response data set comprising a plurality of user queries and a plurality of answers obtained from the large language model for each of the plurality of user queries from an answer generation system that generates answers using a large language model; A step of generating multiple posts using the above user question-response data set; A step of providing a content feed including the plurality of posts to a user terminal; and A program characterized by comprising commands for performing a step of providing different information related to the user query response data set to the user terminal based on different types of user input to the user terminal.

15. A method for generating a content feed using the answers of a large language model, A step of receiving a user query from a user terminal; A step of inputting a prompt including the user query into the large language model and obtaining an answer corresponding to the user query from the large language model; A step of providing the above answer to the user terminal; A step of receiving a sharing request for the user query and the answer from the user terminal; and A method for generating a content feed using answers from a large language model, characterized in that the method comprises the step of providing a user question-response data set including the user question and the answer to a content feed providing system in response to the sharing request.

16. In paragraph 15, A method for generating a content feed using answers from a large language model, further comprising the step of providing a content feed consisting of posts including the user query and the answer to users accessing the content feed providing system.

Images